xmed-lab / diffcmr Goto Github PK

Tianqi Xiang*, Wenjun Yue*, Yiqun Lin, Jiewen Yang, Zhenkun Wang, Xiaomeng Li, "DiffCMR: Fast Cardiac MRI Reconstruction with Diffusion Probabilistic Models" MICCAI2023 CMRxRecon Workshop

Download the CMRxRecon dataset from the official website.

Open data_task1.ipynb and modify 2 variables:

original_dataset_path = "/<your path>/MICCAIChallenge2023/ChallengeData"
save_prepared_dataset_path = "/<your path>/MICCAIChallenge2023/Task1"

Run data_task1.ipynb to get preprocessed data and 6 pair files for Task 1.

Open data_task2.ipynb and modify 3 variables:

dataroot = "/<your path>/MICCAIChallenge2023/ChallengeData"
saveroot = "/<your path>/MICCAIChallenge2023/Task2"
pair_dir = "/<your path>/MICCAIChallenge2023/Task2_pair"

Run data_task2.ipynb to get preprocessed data and pair files for Task 2.

Note that pair files are generated to record the absolute paths of preprocessed data, as long as the path for preprocessed data remains unchanged, you can relocate the pair files at your convience.

Open train_task1_diff.py, modify the variable trainpairfile to the path of pair file for Task1 AccFactor 04/08/10.

python train_task1_diff.py

Open train_task2_diff.py, modify the variable trainpairfile to the path of pair file for Task2 AccFactor 04/08/10.

python train_task2_diff.py

After training, model weights are stored in logdir set in the training files. Note that you will be running 6 times to conduct the entire experiment for both tasks and all coil numbers.

Open inferece.py, modify 3 variables:

result_dir = "<path to store results>"
model_path = "<path of model weights>"
val_pair_file = "<validation pair file>"

Execute inferece.py. Results are listed in the result_dir.